Re: [PATCH] ocfs2: remove unreasonable unlock

From: heming.zhao@xxxxxxxx
Date: Tue Aug 20 2024 - 00:04:58 EST




On 8/19/24 10:51, Lizhi Xu wrote:
There was a lock release before exiting, so remove the unreasonable unlock.

Reported-and-tested-by: syzbot+ab134185af9ef88dfed5@xxxxxxxxxxxxxxxxxxxxxxxxx
Closes: https://syzkaller.appspot.com/bug?extid=ab134185af9ef88dfed5
Signed-off-by: Lizhi Xu <lizhi.xu@xxxxxxxxxxxxx>
---
fs/ocfs2/buffer_head_io.c | 1 -
1 file changed, 1 deletion(-)

diff --git a/fs/ocfs2/buffer_head_io.c b/fs/ocfs2/buffer_head_io.c
index cdb9b9bdea1f..e62c7e1de4eb 100644
--- a/fs/ocfs2/buffer_head_io.c
+++ b/fs/ocfs2/buffer_head_io.c
@@ -235,7 +235,6 @@ int ocfs2_read_blocks(struct ocfs2_caching_info *ci, u64 block, int nr,
if (bhs[i] == NULL) {
bhs[i] = sb_getblk(sb, block++);
if (bhs[i] == NULL) {
- ocfs2_metadata_cache_io_unlock(ci);
status = -ENOMEM;
mlog_errno(status);
/* Don't forget to put previous bh! */

The fix looks good to me. However, I found another issue in this function.

In the for-loop after the 'read_failure' label, the condition
'(bh == NULL) && flags includes OCFS2_BH_READAHEAD' is missing.
When this contidion is true, this for-loop will call ocfs2_set_buffer_uptodate(ci, bh),
which then triggers a NULL pointer access error.

If you agree with my analysis, please add the new fix and send a v2 patch.

Thanks,
Heming